alcohol <- read.csv("student-mat-recoded.csv")
head(alcohol)
## school sex age address famsize Pstatus Medu
## 1 GP F 18 Urban GT3 alone higher education
## 2 GP F 17 Urban GT3 together 4th grade
## 3 GP F 15 Urban LE3 together 4th grade
## 4 GP F 15 Urban GT3 together higher education
## 5 GP F 16 Urban GT3 together secondary education
## 6 GP M 16 Urban LE3 together higher education
## Fedu Mjob Fjob reason guardian traveltime
## 1 higher education at_home teacher course mother 15-30 min
## 2 4th grade at_home other course father <15 min
## 3 4th grade at_home other other mother <15 min
## 4 5th - 9th grade health services home mother <15 min
## 5 secondary education other other home father <15 min
## 6 secondary education services other reputation mother <15 min
## studytime failures schoolsup famsup paid activities nursery higher internet
## 1 2-5 hrs 0 yes no no no yes yes no
## 2 2-5 hrs 0 no yes no no no yes yes
## 3 2-5 hrs 3 yes no yes no yes yes yes
## 4 5-10 hrs 0 no yes yes yes yes yes yes
## 5 2-5 hrs 0 no yes yes no yes yes no
## 6 2-5 hrs 0 no yes yes yes yes yes yes
## romantic famrel freetime goout Dalc Walc health absences G1
## 1 no good moderate high very low very low moderate 6 5
## 2 no excellent moderate moderate very low very low moderate 4 5
## 3 no good moderate low low moderate moderate 10 7
## 4 yes moderate low low very low very low very good 2 15
## 5 no good moderate low very low low very good 4 6
## 6 no excellent high low very low low very good 10 15
## G2 G3
## 1 6 6
## 2 5 6
## 3 8 10
## 4 14 15
## 5 10 10
## 6 15 15
run 4 lasso regressions separated on 4 types of variables: Personal, Main Academic, Family, and School Variables
General Interpretation: For each Walc prediction, the coefficient shown is related to whether it is associated with the Walc outcome and how much it affects the outcome.
#Transform data into matrix
x1 <- model.matrix( Walc ~ 1 + sex + age + address + internet + romantic + health, alcohol)[,-1]
y1 <- as.matrix(alcohol["Walc"])
grid=10^seq(10,-2, length =100)
lasso.mod <- glmnet(x1,y1,alpha = 1, lambda = grid, family="multinomial")
plot(lasso.mod)
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
set.seed(123)
#Split data into train, test dataset
train <- sample(1:nrow(x1),nrow(x1)/2)
test <- (-train)
# Run Cross-Validatiom
cv.out <- cv.glmnet(x1,y1,alpha = 1, family="multinomial")
plot(cv.out)
best_lambda <- cv.out$lambda.min
best_lambda
## [1] 0.0200501
lasso.coef <- predict (lasso.mod ,type = "coefficients",s= best_lambda)
lasso.coef
## $high
## 10 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -1.24620344
## sexM 0.79275873
## age .
## addressUrban .
## internetyes .
## romanticyes -0.02327201
## healthgood .
## healthmoderate .
## healthvery bad .
## healthvery good 0.34617525
##
## $low
## 10 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -0.1647016
## sexM .
## age .
## addressUrban .
## internetyes .
## romanticyes .
## healthgood .
## healthmoderate .
## healthvery bad .
## healthvery good .
##
## $moderate
## 10 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -0.253329589
## sexM .
## age 0.001662019
## addressUrban .
## internetyes .
## romanticyes .
## healthgood .
## healthmoderate .
## healthvery bad .
## healthvery good .
##
## $`very high`
## 10 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -1.9735655
## sexM 1.1606924
## age .
## addressUrban .
## internetyes .
## romanticyes .
## healthgood 0.1102719
## healthmoderate .
## healthvery bad .
## healthvery good .
##
## $`very low`
## 10 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 3.637800160
## sexM -0.153129376
## age -0.196731895
## addressUrban 0.266810422
## internetyes -0.098800380
## romanticyes .
## healthgood 0.008123843
## healthmoderate 0.119964233
## healthvery bad .
## healthvery good -0.121547057
Interpretation: Because the variable internet appears the least in all 5 outcomes, it seems that it might not have a huge effect on alcohol consumption.
x1 <- model.matrix( Walc ~ 1 + studytime + failures + schoolsup + paid + nursery + higher + G1 + G2 + G3, alcohol)[,-1]
y1 <- as.matrix(alcohol["Walc"])
grid=10^seq(10,-2, length =100)
lasso.mod <- glmnet(x1,y1,alpha = 1, lambda = grid, family="multinomial")
plot(lasso.mod)
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
set.seed(123)
#Split data into train, test dataset
train <- sample(1:nrow(x1),nrow(x1)/2)
test <- (-train)
# Run Cross-Validatiom
cv.out <- cv.glmnet(x1,y1,alpha = 1, family="multinomial")
plot(cv.out)
best_lambda <- cv.out$lambda.min
best_lambda
## [1] 0.01225056
lasso.coef <- predict (lasso.mod ,type = "coefficients",s= best_lambda)
lasso.coef
## $high
## 12 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 0.25033443
## studytime>10 hrs -0.11470444
## studytime2-5 hrs .
## studytime5-10 hrs -0.82049409
## failures .
## schoolsupyes .
## paidyes .
## nurseryyes .
## higheryes .
## G1 -0.03719768
## G2 -0.01727566
## G3 .
##
## $low
## 12 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -0.29810238
## studytime>10 hrs .
## studytime2-5 hrs .
## studytime5-10 hrs .
## failures .
## schoolsupyes -0.52301411
## paidyes .
## nurseryyes .
## higheryes 0.38046745
## G1 0.04653959
## G2 .
## G3 -0.04276966
##
## $moderate
## 12 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 0.00370346
## studytime>10 hrs .
## studytime2-5 hrs 0.01614941
## studytime5-10 hrs .
## failures .
## schoolsupyes .
## paidyes 0.02924976
## nurseryyes .
## higheryes .
## G1 .
## G2 .
## G3 .
##
## $`very high`
## 12 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -0.1510636
## studytime>10 hrs .
## studytime2-5 hrs .
## studytime5-10 hrs -0.4148292
## failures 0.2753324
## schoolsupyes .
## paidyes .
## nurseryyes -0.3072604
## higheryes -0.7802255
## G1 .
## G2 .
## G3 .
##
## $`very low`
## 12 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 0.195128101
## studytime>10 hrs 1.289378128
## studytime2-5 hrs 0.475366568
## studytime5-10 hrs 0.761156257
## failures -0.070488118
## schoolsupyes 0.546655950
## paidyes -0.530494334
## nurseryyes 0.173155369
## higheryes -0.018493151
## G1 0.001243482
## G2 0.003967575
## G3 .
Interpretation: BEcause all variables appear about the same time (2 - 4 times) in each Walc observation, all the variables might be associated to the Walc outcome.
x1 <- model.matrix( Walc ~ 1 + famsize + Pstatus + Medu + Fedu + Mjob + Fjob + guardian + famsup + famrel, alcohol)[,-1]
y1 <- as.matrix(alcohol["Walc"])
grid=10^seq(10,-2, length =100)
lasso.mod <- glmnet(x1,y1,alpha = 1, lambda = grid, family="multinomial")
plot(lasso.mod)
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
set.seed(123)
#Split data into train, test dataset
train <- sample(1:nrow(x1),nrow(x1)/2)
test <- (-train)
# Run Cross-Validatiom
cv.out <- cv.glmnet(x1,y1,alpha = 1, family="multinomial")
plot(cv.out)
best_lambda <- cv.out$lambda.min
best_lambda
## [1] 0.05000156
lasso.coef <- predict (lasso.mod ,type = "coefficients",s= best_lambda)
lasso.coef
## $high
## 26 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -0.2857815
## famsizeLE3 .
## Pstatustogether .
## Medu5th - 9th grade .
## Meduhigher education .
## Medunone .
## Medusecondary education .
## Fedu5th - 9th grade .
## Feduhigher education .
## Fedunone .
## Fedusecondary education .
## Mjobhealth .
## Mjobother .
## Mjobservices .
## Mjobteacher .
## Fjobhealth .
## Fjobother .
## Fjobservices .
## Fjobteacher .
## guardianmother .
## guardianother .
## famsupyes .
## famrelexcellent .
## famrelgood .
## famrelmoderate .
## famrelvery bad .
##
## $low
## 26 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 0.2250439
## famsizeLE3 .
## Pstatustogether .
## Medu5th - 9th grade .
## Meduhigher education .
## Medunone .
## Medusecondary education .
## Fedu5th - 9th grade .
## Feduhigher education .
## Fedunone .
## Fedusecondary education .
## Mjobhealth .
## Mjobother .
## Mjobservices .
## Mjobteacher .
## Fjobhealth .
## Fjobother .
## Fjobservices .
## Fjobteacher .
## guardianmother .
## guardianother .
## famsupyes .
## famrelexcellent .
## famrelgood .
## famrelmoderate .
## famrelvery bad .
##
## $moderate
## 26 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 0.14724831
## famsizeLE3 .
## Pstatustogether 0.01939514
## Medu5th - 9th grade .
## Meduhigher education .
## Medunone .
## Medusecondary education .
## Fedu5th - 9th grade .
## Feduhigher education .
## Fedunone .
## Fedusecondary education .
## Mjobhealth .
## Mjobother .
## Mjobservices .
## Mjobteacher .
## Fjobhealth -0.00596914
## Fjobother .
## Fjobservices .
## Fjobteacher .
## guardianmother .
## guardianother .
## famsupyes .
## famrelexcellent .
## famrelgood .
## famrelmoderate .
## famrelvery bad .
##
## $`very high`
## 26 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -0.8854029
## famsizeLE3 .
## Pstatustogether .
## Medu5th - 9th grade .
## Meduhigher education .
## Medunone .
## Medusecondary education .
## Fedu5th - 9th grade .
## Feduhigher education .
## Fedunone .
## Fedusecondary education .
## Mjobhealth .
## Mjobother .
## Mjobservices .
## Mjobteacher .
## Fjobhealth .
## Fjobother .
## Fjobservices .
## Fjobteacher .
## guardianmother .
## guardianother .
## famsupyes .
## famrelexcellent .
## famrelgood .
## famrelmoderate .
## famrelvery bad .
##
## $`very low`
## 26 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 0.7988921401
## famsizeLE3 -0.0055215615
## Pstatustogether .
## Medu5th - 9th grade .
## Meduhigher education .
## Medunone .
## Medusecondary education .
## Fedu5th - 9th grade .
## Feduhigher education .
## Fedunone 0.0576033117
## Fedusecondary education .
## Mjobhealth .
## Mjobother .
## Mjobservices .
## Mjobteacher .
## Fjobhealth 0.0480260141
## Fjobother .
## Fjobservices .
## Fjobteacher .
## guardianmother .
## guardianother .
## famsupyes .
## famrelexcellent .
## famrelgood .
## famrelmoderate -0.0009581581
## famrelvery bad .
Interpretation: For most of Walc outcome, it seems that most variables to not appear, which might mean that generally, the variables within the Family category might not be associated to the Walc outcome.
x1 <- model.matrix( Walc ~ 1 + school + reason + traveltime + activities + freetime + goout + absences, alcohol)[,-1]
y1 <- as.matrix(alcohol["Walc"])
grid=10^seq(10,-2, length =100)
lasso.mod <- glmnet(x1,y1,alpha = 1, lambda = grid, family="multinomial")
plot(lasso.mod)
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
set.seed(123)
#Split data into train, test dataset
train <- sample(1:nrow(x1),nrow(x1)/2)
test <- (-train)
# Run Cross-Validatiom
cv.out <- cv.glmnet(x1,y1,alpha = 1, family="multinomial")
plot(cv.out)
best_lambda <- cv.out$lambda.min
best_lambda
## [1] 0.01667076
lasso.coef <- predict (lasso.mod ,type = "coefficients",s= best_lambda)
lasso.coef
## $high
## 18 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 0.23404143
## schoolMS .
## reasonhome .
## reasonother .
## reasonreputation -0.12716076
## traveltime>1 hr .
## traveltime15-30 min .
## traveltime30 min - 1 hr 0.01961985
## activitiesyes .
## freetimelow .
## freetimemoderate -0.05541240
## freetimevery high .
## freetimevery low -0.57397535
## gooutlow -0.29502578
## gooutmoderate -1.05070386
## gooutvery high 0.12535695
## gooutvery low .
## absences 0.01419382
##
## $low
## 18 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 0.022786562
## schoolMS .
## reasonhome .
## reasonother -0.033837479
## reasonreputation .
## traveltime>1 hr .
## traveltime15-30 min .
## traveltime30 min - 1 hr .
## activitiesyes -0.029498862
## freetimelow 1.000133206
## freetimemoderate 0.007374884
## freetimevery high .
## freetimevery low .
## gooutlow 0.755763462
## gooutmoderate 0.329836537
## gooutvery high .
## gooutvery low 0.026561956
## absences -0.015604173
##
## $moderate
## 18 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 0.2764548
## schoolMS 0.4214981
## reasonhome .
## reasonother .
## reasonreputation 0.1069719
## traveltime>1 hr .
## traveltime15-30 min .
## traveltime30 min - 1 hr .
## activitiesyes .
## freetimelow .
## freetimemoderate .
## freetimevery high .
## freetimevery low .
## gooutlow .
## gooutmoderate .
## gooutvery high .
## gooutvery low .
## absences .
##
## $`very high`
## 18 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -1.2811584466
## schoolMS .
## reasonhome .
## reasonother 0.0059139055
## reasonreputation .
## traveltime>1 hr 2.7284452349
## traveltime15-30 min -0.0602850283
## traveltime30 min - 1 hr .
## activitiesyes .
## freetimelow .
## freetimemoderate .
## freetimevery high .
## freetimevery low 0.0311984372
## gooutlow .
## gooutmoderate -0.0001389792
## gooutvery high 1.6377032021
## gooutvery low .
## absences .
##
## $`very low`
## 18 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 0.74787564
## schoolMS -0.36943466
## reasonhome .
## reasonother -0.21794296
## reasonreputation .
## traveltime>1 hr .
## traveltime15-30 min .
## traveltime30 min - 1 hr -0.41299230
## activitiesyes 0.26050677
## freetimelow .
## freetimemoderate .
## freetimevery high -0.06830581
## freetimevery low 0.36409196
## gooutlow 0.91492105
## gooutmoderate .
## gooutvery high -0.24073372
## gooutvery low 1.15008333
## absences -0.01620053
Interpretation: Generally, it seems that the variables are consistently appearing for each Walc observation, which might mean that the variables within this category have association with the Walc outcome.
x1 <- model.matrix( Walc ~ 1 + sex + age + address + internet + romantic + health + studytime + failures + schoolsup + paid + nursery + higher + G1 + G2 + G3 + famsize + Pstatus + Medu + Fedu + Mjob + Fjob + guardian + famsup + famrel + school + reason + traveltime + activities + freetime + goout + absences, alcohol)[,-1]
y1 <- as.matrix(alcohol["Walc"])
grid=10^seq(10,-2, length =100)
lasso.mod <- glmnet(x1,y1,alpha = 1, lambda = grid, family="multinomial")
plot(lasso.mod)
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
## Warning in regularize.values(x, y, ties, missing(ties), na.rm = na.rm):
## collapsing to unique 'x' values
set.seed(123)
#Split data into train, test dataset
train <- sample(1:nrow(x1),nrow(x1)/2)
test <- (-train)
# Run Cross-Validatiom
cv.out <- cv.glmnet(x1,y1,alpha = 1, family="multinomial")
plot(cv.out)
best_lambda <- cv.out$lambda.min
best_lambda
## [1] 0.02203776
lasso.coef <- predict (lasso.mod ,type = "coefficients",s= best_lambda)
lasso.coef
## $high
## 63 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -0.1284263039
## sexM 0.7667975886
## age .
## addressUrban .
## internetyes .
## romanticyes -0.0046782999
## healthgood -0.0002491843
## healthmoderate .
## healthvery bad .
## healthvery good 0.2806202242
## studytime>10 hrs .
## studytime2-5 hrs .
## studytime5-10 hrs -0.0153482923
## failures .
## schoolsupyes .
## paidyes .
## nurseryyes .
## higheryes .
## G1 -0.0238068802
## G2 .
## G3 .
## famsizeLE3 .
## Pstatustogether .
## Medu5th - 9th grade -0.1570666514
## Meduhigher education .
## Medunone 0.0378641745
## Medusecondary education 0.1233823278
## Fedu5th - 9th grade .
## Feduhigher education .
## Fedunone .
## Fedusecondary education 0.0060123576
## Mjobhealth .
## Mjobother .
## Mjobservices .
## Mjobteacher .
## Fjobhealth .
## Fjobother .
## Fjobservices .
## Fjobteacher -0.0800347072
## guardianmother .
## guardianother -0.4085083436
## famsupyes -0.1060857850
## famrelexcellent .
## famrelgood .
## famrelmoderate 0.0213395012
## famrelvery bad .
## schoolMS .
## reasonhome .
## reasonother .
## reasonreputation .
## traveltime>1 hr .
## traveltime15-30 min .
## traveltime30 min - 1 hr .
## activitiesyes .
## freetimelow .
## freetimemoderate .
## freetimevery high .
## freetimevery low -0.1075673746
## gooutlow -0.2256195249
## gooutmoderate -0.8538818529
## gooutvery high 0.0248931099
## gooutvery low .
## absences 0.0102634017
##
## $low
## 63 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -0.039792125
## sexM .
## age .
## addressUrban .
## internetyes 0.017018142
## romanticyes .
## healthgood .
## healthmoderate .
## healthvery bad .
## healthvery good .
## studytime>10 hrs .
## studytime2-5 hrs .
## studytime5-10 hrs .
## failures .
## schoolsupyes -0.325712940
## paidyes .
## nurseryyes .
## higheryes .
## G1 .
## G2 .
## G3 -0.007738140
## famsizeLE3 .
## Pstatustogether 0.109419890
## Medu5th - 9th grade .
## Meduhigher education .
## Medunone -0.071583392
## Medusecondary education .
## Fedu5th - 9th grade .
## Feduhigher education 0.004439740
## Fedunone .
## Fedusecondary education .
## Mjobhealth -0.109152912
## Mjobother .
## Mjobservices .
## Mjobteacher .
## Fjobhealth .
## Fjobother .
## Fjobservices .
## Fjobteacher 0.007902992
## guardianmother .
## guardianother .
## famsupyes .
## famrelexcellent 0.008406682
## famrelgood .
## famrelmoderate .
## famrelvery bad .
## schoolMS .
## reasonhome .
## reasonother .
## reasonreputation .
## traveltime>1 hr .
## traveltime15-30 min .
## traveltime30 min - 1 hr .
## activitiesyes .
## freetimelow 0.907403942
## freetimemoderate .
## freetimevery high .
## freetimevery low .
## gooutlow 0.656287057
## gooutmoderate 0.273033991
## gooutvery high .
## gooutvery low .
## absences -0.011128079
##
## $moderate
## 63 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -0.2566559481
## sexM .
## age .
## addressUrban .
## internetyes .
## romanticyes .
## healthgood .
## healthmoderate .
## healthvery bad .
## healthvery good .
## studytime>10 hrs .
## studytime2-5 hrs .
## studytime5-10 hrs .
## failures .
## schoolsupyes .
## paidyes .
## nurseryyes .
## higheryes .
## G1 .
## G2 .
## G3 0.0001109037
## famsizeLE3 .
## Pstatustogether 0.5437422894
## Medu5th - 9th grade -0.0165928870
## Meduhigher education .
## Medunone .
## Medusecondary education 0.0060769150
## Fedu5th - 9th grade .
## Feduhigher education -0.0916196474
## Fedunone .
## Fedusecondary education .
## Mjobhealth 0.1663051503
## Mjobother .
## Mjobservices .
## Mjobteacher .
## Fjobhealth -0.6188636081
## Fjobother 0.0018459564
## Fjobservices .
## Fjobteacher .
## guardianmother .
## guardianother .
## famsupyes .
## famrelexcellent .
## famrelgood .
## famrelmoderate .
## famrelvery bad .
## schoolMS 0.3146232650
## reasonhome .
## reasonother 0.0007339715
## reasonreputation 0.0397818743
## traveltime>1 hr .
## traveltime15-30 min .
## traveltime30 min - 1 hr .
## activitiesyes .
## freetimelow .
## freetimemoderate .
## freetimevery high .
## freetimevery low .
## gooutlow .
## gooutmoderate .
## gooutvery high .
## gooutvery low .
## absences .
##
## $`very high`
## 63 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) -1.5400480208
## sexM 0.8954974751
## age .
## addressUrban .
## internetyes .
## romanticyes .
## healthgood 0.0221212479
## healthmoderate .
## healthvery bad .
## healthvery good .
## studytime>10 hrs .
## studytime2-5 hrs .
## studytime5-10 hrs .
## failures 0.0002364946
## schoolsupyes .
## paidyes .
## nurseryyes -0.1574068345
## higheryes -0.1735073661
## G1 .
## G2 .
## G3 .
## famsizeLE3 .
## Pstatustogether .
## Medu5th - 9th grade .
## Meduhigher education .
## Medunone .
## Medusecondary education .
## Fedu5th - 9th grade .
## Feduhigher education .
## Fedunone .
## Fedusecondary education .
## Mjobhealth .
## Mjobother .
## Mjobservices .
## Mjobteacher .
## Fjobhealth .
## Fjobother .
## Fjobservices .
## Fjobteacher .
## guardianmother .
## guardianother .
## famsupyes .
## famrelexcellent .
## famrelgood .
## famrelmoderate .
## famrelvery bad 0.0055536703
## schoolMS .
## reasonhome .
## reasonother .
## reasonreputation .
## traveltime>1 hr 2.4452923989
## traveltime15-30 min -0.0112986965
## traveltime30 min - 1 hr .
## activitiesyes .
## freetimelow .
## freetimemoderate .
## freetimevery high .
## freetimevery low .
## gooutlow .
## gooutmoderate .
## gooutvery high 1.4708657241
## gooutvery low .
## absences .
##
## $`very low`
## 63 x 1 sparse Matrix of class "dgCMatrix"
## 1
## (Intercept) 1.964922398
## sexM -0.142256301
## age -0.095531035
## addressUrban 0.378305666
## internetyes .
## romanticyes .
## healthgood .
## healthmoderate 0.017202315
## healthvery bad .
## healthvery good -0.175358754
## studytime>10 hrs 0.715322837
## studytime2-5 hrs 0.016818753
## studytime5-10 hrs 0.305079385
## failures .
## schoolsupyes 0.154069473
## paidyes -0.394455682
## nurseryyes 0.103816309
## higheryes -0.004508234
## G1 .
## G2 .
## G3 .
## famsizeLE3 -0.113005044
## Pstatustogether .
## Medu5th - 9th grade .
## Meduhigher education .
## Medunone .
## Medusecondary education .
## Fedu5th - 9th grade .
## Feduhigher education .
## Fedunone 0.683562042
## Fedusecondary education .
## Mjobhealth .
## Mjobother .
## Mjobservices .
## Mjobteacher .
## Fjobhealth 0.043791214
## Fjobother .
## Fjobservices -0.161007737
## Fjobteacher 0.116485556
## guardianmother .
## guardianother 0.001157082
## famsupyes 0.015332479
## famrelexcellent 0.186640603
## famrelgood .
## famrelmoderate -0.306438167
## famrelvery bad -0.045356821
## schoolMS .
## reasonhome .
## reasonother -0.064126754
## reasonreputation .
## traveltime>1 hr .
## traveltime15-30 min 0.021089547
## traveltime30 min - 1 hr -0.206855983
## activitiesyes 0.240127239
## freetimelow .
## freetimemoderate .
## freetimevery high .
## freetimevery low 0.269798668
## gooutlow 0.912883037
## gooutmoderate .
## gooutvery high -0.163580324
## gooutvery low 1.015974795
## absences -0.007347194